An Event-Based Approach for Model-Based Control and Parameter Identification in Networked Distributed Processes

• Published in: Proceedings of the American Control Conference pp. 3425 - 3430
• Main Authors: Zedan, Amr, El-Farra, Nael H.
• Format: Conference Proceeding
• Language: English
• Published: AACC 01.07.2020
• Subjects: Eigenvalues and eigenfunctions; Estimation error; Feedback control; Mathematical model; Process control; Stability analysis; Uncertainty
• ISSN: 2378-5861
• DOI: 10.23919/ACC45564.2020.9147741

This work focuses on the problem of resource-constrained stabilization of spatially-distributed systems modeled by PDEs with low-order dynamics, subject to sensor-controller communication constraints and process parametric variations. An approach that brings together event-triggered model-based control and event-based parameter re-identification is developed to maintain closed-loop stability in the presence of parametric drift, while simultaneously limiting the rate of sensor-to-controller information transfer. Initially, a model-based feedback controller with an event-triggered sensor-controller communication logic is designed on the basis of an approximate finite-dimensional model, and its implementation on the infinite-dimensional system is investigated. An event-based parameter re-identification and update strategy is incorporated within the model-based control strategy to avert the need for a permanent increase in the sensor-controller communication rate in response to the destabilizing influence of process drift. A key component of this strategy is the design of a moving-horizon communication frequency monitoring scheme that detects sustained increases in post-drift communication and triggers parameter re-identification whenever a certain model state update frequency threshold is breached. The close-dloop stability and communication requirements associated with the newly-identified model are analyzed and used to decide whether to update the model parameters. In the event of parameter updates, a new closed-loop stability threshold is obtained based on the updated model to trigger future sensor-controller communications appropriately. The development and implementation of the proposed approach are illustrated using a representative diffusion-reaction process example.